Journal of Structural Engineering

This paper presents an investigation on the variation of ductility capacity of steel beam‐columns. First, experimental data obtained from previous tests are surveyed, and their ductility ratios are examined. It is found that experimental ductility ratios scatter over a wide range with their coefficients of variation ranging from 0.4 to 0.9. Next, nonlinear numerical analysis is carried out with slenderness, axial force, strain hardening, residual stresses, and initial out‐of‐straightness as variables. The analysis reveals that neither residual stresses nor initial out‐of‐straightness is a major factor affecting the ductility ratio, but strain hardening is very influential on the ductility ratio, suggesting that strain hardening is one of the major sources for the large variation in the experimental ductility ratio. A guideline is also presented for estimating ductility ratios of steel beam‐columns. Finally, problems associated with large variation in the ductility capacity of steel beam‐columns are discussed with reference to the reliability design format.

Structural‐parameter identification of member properties of framed structures is conducted at the element level. By substructuring, a frame element is isolated from the complete structure, and the state equation used for parameter estimation is derived from the equation of motion of the frame element. Using the extended Kalman filter with weighted global iteration, the physical characteristics such as the axial rigidity, flexural rigidity, and damping parameters of a frame element are identified. Numerical simulation studies are done for two‐story plane frames, and the member properties of the frame elements are estimated. The convergence behavior of the parameters is observed and the possible cause of divergence is explored. The results show that the convergence behavior of the identified parameters is affected by a number of factors such as initial guess of the parameters, number of observed responses, and the choice of observed response.

This paper discusses the transmitting capacity of the so‐called surface or metal contact joints by which the compressive stresses are transmitted through the directly contacting surface of steel plates. The imperfections of the contacted surface are modeled by the cylindrical and periodical sinusoidal form, and the stress distribution up to the yield state is simulated using the finite element method as a contact problem. The yielded portion of the plate spreads in the relatively small portion close to the surface, and the longitudinal displacements are not so large in this state. Therefore, the contact joint can transmit compressive loads of almost the full plastic capacity. The stress sharing with the splice plate is also evaluated by a simple spring model associated with the roughness of the contact joint, and is closely related to the bolt pitch. Therefore, the maximum tolerance for the gaps should be determined by the tolerable deformation of the contact joints and the requirements for the combined bolted joints as a total system.

Corrosion fatigue is a complex phenomenon involving many variables. Welded‐steel tubular joints of offshore structures are subjected to corrosion fatigue due to the combined action of corrosive seawater environment and cyclic wave forces. This paper addresses the aforementioned problem, mainly with respect to internally ring stiffened steel tubular joints. The studies were made with respect to tropical seawater environment similar to that in the Bombay High offshore region where bulk of the Indian oil is produced. Corrosion‐fatigue tests were conducted on five welded‐steel tubular T and Y joints under constant amplitude axial‐brace loading at a frequency of 0.2 Hz. The studies have shown reduction in fatigue life when compared to S‐N curves recommended by the codes that are mainly based on air fatigue tests. Based on crack‐growth data obtained from the tests, a model has been proposed to predict the remaining life of stiffened tubular joints under corrosion fatigue.

Flexible aluminum panels of various thickness subjected to blast loads have been analyzed by different techniques; an equivalent single degree of freedom program using the DUHAMEL integration technique and a multiple‐degree‐of‐freedom finite element program. Panels of various thickness have been modeled as simply supported plates using equivalent beams with moment of inertia adjusted to account for the greater stiffness of plate action. Two dynamic load types were examined. The first type correctly depicts nonlinear blast loads. The second type, a triangular load, is a commonly employed approximation of nonlinear blast pressures. Comparisons of dynamic reduction factors for the structural response shows that the dynamic nature of blast loads reduce panel response significantly. Frequently used triangular pressure distribution for simulation of blast pressure, is found to be overly conservative. A better match is obtained by reducing the duration of triangular pressure pulse. An equivalent pulse duration is provided.

Steel bridge piers under cyclic loading due to strong earthquake motion should be designed considering not only the strength but also the ductility capacity. Steel sections with concrete infill are being increasingly used as structural members, since filling the steel section with concrete results in an increase in both strength and ductility without increasing the section size. In this paper, the results of 11 cantilever thin‐walled steel box column specimens tested under constant compressive axial loads and cyclic lateral loads are presented. The test specimens were of unstiffened and stiffened box sections. To examine the possibility of increasing the column's ductility capacity, seven partially concrete‐filled column specimens were tested. The test results were evaluated using both energy‐absorption capacity and a newly defined ductility factor that takes into account the effect of plate local buckling. It has been shown that the concrete‐filled columns increase both ductility and energy‐absorption capacity significantly.

This paper presents a nonlinear inelastic analysis of the biaxial bending and torsion of thin‐walled steel beam‐columns based on the principle of virtual work. The effect of geometric nonlinearity is developed using position vector analysis. Approximations are not made in the early stage of the development, and thus some significant terms for buckling and postbuckling analysis are retained. The von Mises yield criterion, the associated flow rule, and the hardening rule are used in formulating the elastic‐plastic constitutive matrix for the material inelasticity. Inelastic uniform and nonuniform torsion are incorporated in the formulation as well as the linear and nonlinear geometric effects of the loads. A corresponding finite‐element model for the nonlinear incremental analysis of biaxial bending and torsion of thin‐walled beam‐columns is presented by using the principle of virtual work. Numerical results in a companion paper indicate that the model presented is capable of making accurate nonlinear analyses including postbuckling analyses.

In a companion paper, a nonlinear inelastic analysis of the biaxial bending and torsion of beam‐columns was presented. Approximations were not made in the early stage of the development, and thus some significant terms were retained. The present paper uses a finite‐element procedure of the companion paper to investigate numerically the elastic‐plastic behavior of I‐beams. The problems include the large‐deflection in‐plane elastic and inelastic behavior of columns and beam‐columns; torsion of beams and beam‐columns; eccentric and inclined loading of beams; the effects of initial crookedness and twist on the lateral buckling strength of beams; and the postbuckling flexural‐torsional behavior of beams. The flexural‐torsional design strength of steel I‐beams is also studied. The results indicate that the procedure is capable of carrying out an accurate nonlinear analysis including postbuckling analysis. It is demonstrated that ignoring some significant terms may lead to an overestimation of buckling loads, poor prediction of post‐buckling behavior, and failure to find the correct maximum load capacity.

A large‐scale experimental test program was conducted to assess the residual strength and repair of dent‐damaged offshore tubular bracing. The program involved performing 13 tests that included nondamaged, dent‐damaged, and repaired specimens. Damaged specimens had a dent depth of 10% of their diameter. The nominal diameter‐to‐thickness $(D/t)$ ratios were 34.5, 46, and 64 for the specimens. The dent damage significantly reduced the specimen's strength, indicating the need for repair. The repair of the dented specimens involved using internal grouting and a grouted steel sleeve. Test results showed that by arresting the dent‐depth growth, both repair techniques were able to reinstate a damaged specimen's strength. Existing and newly developed ultimate‐strength formulations for damaged members, and existing strength formulations for repaired members were evaluated through a comparison with specimen behavior. Results of the evaluation show a reasonable agreement with test results.

A method for predicting the sectional strength variability of complex concrete structures subjected to combined membrane forces, bending moments, and transverse shear is presented. SHELL474, which uses a three‐dimensional generalization of the modified compression field theory to predict the theoretical capacity of complex concrete elements, is incorporated within the reliability‐analysis model RELAN, in order to evaluate the function defining the failure limit state. The method is used in an example where the strength variability of the 1.4‐m‐thick ice wall of the Hibernia concrete offshore structure (1986 updated design) is studied. Statistical models for the 15 random variables that affect the resistance of the structural concrete wall were determined from the literature. The analysis results illustrate the importance of obtaining updated statistical information on modern high‐performance concrete, for both the compressive and tensile‐strength distributions.

This paper presents the results of a study of rates of deterioration of steel and prestressed concrete bridge superstructures. The study involved forming sample groups of bridges that were homogeneous in terms of characteristics, other than superstructure material, which were expected to influence rates of deterioration. Superstructure condition rating was regressed against age and average daily traffic for sample groups of homogeneous steel and prestressed concrete bridges in the nation as a whole, as well as in seven states selected to represent diverse terrains and environments. Results indicated that age is the primary determinant of superstructure deterioration, followed by average daily traffic. Choice of structural material does not significantly affect superstructure deterioration rates. The study also found that deterioration of bridge superstructures takes the form of a convex function over time, with the rate of observed deterioration slowing as a bridge ages.

An equivalent frame approach is presented for nonlinear seismic analysis of reinforced‐concrete flat plate buildings. The approach employs a parametric hysteretic model and is based on the effective slab‐width concept. Unlike in the previous equivalent frame approaches, the proposed method targets both the moment‐transfer capacity as well as stiffness of the interior and exterior slab‐column connections. The hysteretic parameters and the effective slab width factors are determined from results of laboratory tests on slab‐column connections for the class of flat plate buildings constructed prior to the 1960s with reinforcing detail typical of gravity load design. The validity of the approach is justified by comparing the calculated and the measured responses of two‐bay flat plate subassemblies tested under earthquake‐type loading. With the proposed equivalent frame approach, the response of flat plate buildings for seismic loading could be predicted more realistically over a wide range of lateral drift levels.

The stability behavior of rigid and semirigid frames bent in the weak axis is investigated using a hardening plastic hinge method. The stability limit load of frames is usually overpredicted by the simple plastic hinge analysis neglecting the effect of partial plastification. The plastic zone analysis can predict the limit load more accurately but a large number of elements must be used. Hence, it is costly and time‐consuming. An alternative method called the “hardening plastic hinge method” is developed here for practical use. In the proposed method, one element is used per beam‐column. The initial yield surface including residual stress is first determined from the normalized moment‐curvature‐thrust relationship of $W8\times 31$ section. The member stiffness is then modified to simulate the effect of partial plastification through the concept of work‐hardening in which the average degradation of tangent stiffness of a cross section is calibrated against the slopes of the exact moment‐curvature‐thrust relationship. The stability limit loads of planar steel frames predicted by this and several other methods are compared. The results predicted by the hardening‐hinge method are found close to that of the plastic‐zone method and experimental results.

A destructive test was conducted on a decommissioned skew RC slab bridge. This bridge had severely deteriorated shoulder regions, where the top mat reinforcing bars were exposed. The test bridge was loaded using hydraulic cylinders arranged so as to simulate an HS‐20‐44 truck entering the bridge in one lane. The results indicate that the boundary conditions and the geometry of the bridge significantly influence bridge behavior. During the test, changes in the boundary conditions and yielding of the reinforcing bars caused the load path to shift twice. Failure of the slab occurred due to punching shear at a load of 3,200 kN (720 kips), equivalent to the axle load of 22 HS‐20‐44 trucks. Although the deterioration did not appear to affect flexural behavior, it did greatly reduce the punching‐shear capacity. Analysis indicates that simple modeling methods, such as strip models, overestimate structural demand and underestimate structural supply because they do not properly account for additional load‐carrying mechanisms, such as slab action. Better estimates of structural supply and demand are found using nonlinear finite‐element models.

The static and dynamic loads applied to hand‐set slab formworks during the pouring of a concrete slab are investigated. The spatial distribution and time history of the concrete‐placement loads and the resulting formwork response depend on the equipment used for placing the concrete. The impact loads associated with the crane‐and‐bucket method of concrete placement are investigated. A number of factors affect the magnitude of the concrete load when emptying a bucket. These factors include the height of fall, concrete slump, bucket‐discharge rate, and the dynamic properties of the slab formwork. A single‐degree‐of‐freedom, variable mass, dynamic model is developed for calculating the concrete‐placement loads by taking into consideration the above‐mentioned factors. The concrete‐flow rate from a bucket is modeled as a lognormal white‐noise random process. To verify the validity of the developed model, several lab experiments were performed using a $2.4m\times 2.4m(8\mathrm{ft}\times 8\mathrm{ft})$ formwork deck. The lab's overhead crane was used to lift the concrete‐filled buckets to various heights above the deck before discharging the concrete. Pressure and displacement transducers were attached to the formwork deck for measuring the load intensity and deck displacements.

Long‐term deflection of statically determinate reinforced concrete beams under service loads is analytically studied. The BP‐2 creep and shrinkage model is adopted and extended to accommodate concrete tensile behavior by assuming a tensile stress‐strain relation. The creep rate is related to the stress in the same manner for both tension and compression. Beam deflection is determined using curvatures computed by assuming that the total strain varies linearly across depth (Bernoulli‐Navier hypothesis). Available test results for 37 beams are compared with deflection predictions by this numerical scheme and by applications of the same model after simplification using effective modulus methods. Further comparisons are made with the 1983 ACI Building Code method and a recent method proposed by Bažant and Oh. The direct implementation of the extended BP‐2 model is observed to be the most accurate. The results demonstrate that a realistic model of creep strains can lead to good estimates of deflection. Furthermore, the results indicate that the contribution of shrinkage to overall deformation can be significant.

Composite steel and concrete beams that incorporate a haunch are prone to splitting failure. This is because the shear connectors have to transfer high concentrations of load into the concrete slab in the region of the haunch where the side cover to the connectors is limited. Experimental and theoretical research that has already been published can be used to determine the splitting resistance for haunches with vertical sides. Experimental tests on stud shear connectors encased in haunches with sloping sides is reported, and the results are used to formulate design rules that can be used to prevent composite beams with sloping haunches from splitting. The results can be used to design composite beams with all shapes of haunches and composite L‐beams with sloping sides. Furthermore, they can be used to design steel and concrete composite beams that incorporate composite slabs made with steel decking when the ribs of the steel decking are parallel to the steel section of the composite beam.